Development and validation of the commitment to social activism scale using the Thurstone scaling procedure

ABSTRACT

This paper centers on social activism, a distinctive form of behavior characterized by escalating actions that vary in degrees of difficulty and commitment. Rather than adopting a Likert scaling approach, we embrace a Thurstone technique to construct a commitment to social activism scale (CSAS). We incorporate focus groups, online survey data, and tests of validity in a multistage approach. The validity of CSAS was examined using convergent (activist identity and activist intention) and discriminant (trait anger and dogmatism) scales. Moving forward, the Thurstone scaling procedure used can be mirrored to improve theoretical work studying escalating behaviors across communication topics.

KEYWORDS:

• Activism behavior
• measurement
• Thurstone scaling procedure

Social activism is a unique form of political behavior in which people engage in intentional and strategic actions to bring about social change, but these behaviors can vary in cost (Poorisat et al., 2019). Social activism is an escalating behavior because it is made up of distinct behaviors that vary in their difficulty and commitment levels (Corning & Myers, 2002; Goodwin & Jasper, 2014). The various behaviors that social activism encompasses should not be treated as equivalent, weighted the same, or believed to be synonyms; yet this is oftentimes how they are measured (Klar & Kasser, 2009; McKeever et al., 2023). For example, planning a protest may involve negotiations with law enforcement, recruiting volunteers, and extensive external communications. Contrastingly, retweeting activists on Twitter involves little time or effort. These differences are significant for those who measure or predict social activism in their scholarship (Turner, 2007; Turner et al., 2020), and need to be addressed operationally. As a test of this suggestion, we adopted and adapted the equal-appearing intervals measurement paradigm developed by Thurstone (1927). To do so, we started with focus group data to develop a list of social activism behaviors and the perceived commitment of each behavior. Next, we conducted a survey that measured the perceived commitment of each behavior for both self-described activists and nonactivists. The result is a validated commitment to social activism using a modified multitrait approach.

Measuring activism

There are a few scales that attempt to gauge activism, but majority of these employ Likert-style measurement. For example, McKeever et al. (2023) developed a multi-dimensional activism measure that broke down behaviors as oppositional, collective, financial, symbolic, online, and dialogic. Unfortunately, although the scale contains a variety of behaviors, it fails to capture the escalating nature of activism behavior.

Commitment to Social Activism

Social activism behavior varies in the difficulty, or perceived commitment, of adoption (Turner, 2007). Moreover, certain nonviolent tactics are generally considered to be more socially constructive relative to violent tactics (Zunes et al., 1999) and are more effective in achieving stated goals (Chenoweth & Stephan, 2011), underscoring the notion that all behaviors are not equivalent. Tilly’s “repertoire of contention” describes a range of tactics that can be adopted by activists (Tilly & Tarrow, 2015, p. 472), including boycotts, petitions, marches, and sit-ins. Contemporary activists engage in activism through the Internet and other social media platforms also known as Internet activism or cyberactivism (Meikle, 2002). Taken together, activist behaviors are not interchangeable, and thus should not be treated as a summative scale (e.g., Likert). Thus, as Thurstone created his classic measurement style to assess the favorability of particular attitudes, we will use this type of scaling to assess the perceived commitment of various activist behaviors.

Others have called out the need for addressing the escalating nature of activism. However, the purpose of prior work (Poorisat et al., 2019) was not to conduct a full-scale validation of that measure. Moreover, their use of a Guttman scale assumes linear escalation (Guttman, 1944; Guttman & Suchman, 1947). That is, someone who is willing to engage in item eight on their list (i.e., participate in a demonstration to raise awareness about homelessness) would be assumed to also engage in items 1–7 (e.g., share an online petition to tell the government to help the homeless). We assert it is unlikely that the distance between any two activist behaviors is equivalent, and suggest a Thurstone method. With Thurstone, we can measure the interval by examining the median and interquartile range of the commitment perceived for each behavior and then create a weight for each item.

The Thurstone method

The Thurstone scaling method was designed with the primary purpose of measuring the intensity of one’s attitudes (Thurstone, 1927). Thurstone scales measure attitude by using a series of agree-disagree statements of various intensities. The statements vary in their favorability toward the attitude; as such, these statements assist in operationalizing not only the respondent’s opinion but the strength of that opinion. Thus, it is a cumulative scale, as opposed to a summative scale (i.e., Guttman, 1944; Likert, 1932). Thurstone scales are explicitly built to be unidimensional. Whereas a Likert scale measuring, say, quality of life might ask about physical, mental, financial, and sexual quality, Thurstone scales only ask about favorability—or here, commitment to activism. That means that every participant responds to the same perceived commitment question repeatedly, with only the object of consideration (i.e., behavior) changing.

Procedurally, Thurstone scales are created by first making a large list of items that reflect the construct of interest. Next, this list of items is given to a group of judges with the instruction to rate each item for its favorability (if measuring an attitude). Using these data, the median and interquartile range are computed. Ultimately, these scores are employed to choose the final scale items and their weight in determining intensity. Here, we can create items that measure the degree of commitment to activism by assessing what activism behaviors participants are willing to engage in by first assessing, and creating weights, to reflect the level of commitment of each behavior. A critical step in scale development is establishing the validity of the scale, which can be done using multi-trait methodology or tests of convergent and divergent validity (Campbell & Stanley, 1963) since more traditional tests (e.g., confirmatory factor analyses) are incompatible with unidimensional Thurstone scales.

Convergent Validity

Convergent validity refers to “the extent that measures of constructs that are theoretically related are actually related” (Crano et al., 2015, p. 68). It is identified by the presence of robust correlations between a measure of a focal construct and other theoretically similar constructs. Social activism intention is a Likert scale (Moskalenko & McCauley, 2009) used for individuals who belong to and identify with an organization. It was developed to differentiate activism from radicalism. Yet, this scale does not assess the fine-tuned differences in the level of commitment people are willing to put toward an issue. Likewise, activist identity (Klar & Kasser, 2009) measures one’s belongingness to a particular activist group. For example, people may adopt the identity of “environmentalist.” We are not suggesting that these constructs are the same as CSAS but rather should converge with CSAS.

Discriminant Validity

Discriminant validity (i.e., divergent validity) ensures that constructs that are conceptually distinct from the phenomenon being studied are, in fact, unrelated (Campbell & Fiske, 1959). Here, we identified two trait constructs that could be correlated with CSAS; however, the correlation should be significantly smaller than those between CSAS and the convergent scales. That is, we should be able to reasonably discriminate these trait constructs.

Turner (2007) argued that social activists are often angry about injustice surrounding a social issue and engage in activism to satisfy their need for redemption. If correct in her assertion, Turner is writing about state anger, anger caused by a specific perceived injustice. Trait anger, however, regards a natural tendency to become angry easily, be quick-tempered, or get annoyed (Spielberger et al., 1983). Trait anger should be equally distributed among activists and nonactivists, and therefore weakly correlated. The second discriminant construct is dogmatism. Rokeach (1954) defined dogmatism as “…(a) a relatively closed cognitive system of beliefs and disbeliefs about reality, (b) organized around a central set of beliefs about absolute authority which, in turn, (c) provides a framework for patterns of intolerance and qualified tolerance towards others” (p. 195). Individuals with high dogmatism may more strongly believe in one side of an issue, but they are not necessarily activists.

Given these arguments, we predict:

H1:

CSAS will be strongly correlated with (a) activism intentions and (b) activist identity, compared to trait anger and dogmatism.

H2:

CSAS will be weakly correlated with (a) trait anger and (b) dogmatism, compared to with activism intention and activist identity.

Finally, it is expected that activists find certain activities (e.g., protesting) to be moderately effortful while nonactivists perceive it to be very effortful. It is critical, then, that this scale is usable in studies assessing the behaviors of either audience. Thus, we also address the influence of activist identity on our scale by exploring activists and nonactivists independently and collectively:

RQ1:

Do activists and nonactivists differ in their commitment to social activism?

Study 1

Method

Design

The purpose of Study 1 was to develop face-valid activism behaviors, based on knowledge of activism, to be assessed in Study 2. Face validity was assessed by the authors of this paper and focus group participants. We started with a list of activism behaviors culled from the literature but wanted to ensure the list was exhaustive using focus groups. Four moderated focus groups, each taking about an hour were conducted on Zoom in March 2021. Total sample size (n₁ = 8, n_{2 =} 8, n_{3 =} 3, n_{4 =} 4) was N = 23. Study design and procedures were deemed exempt by the university’s IRB.

Participants

Participants were mostly white (52.2%), female (52.2%), and ranged from 18 to 23 years of age.¹ Participants were drawn from the students’ subject pool at a large Midwestern university and compensated with course credit.

Procedure

Focus groups were conducted online via Zoom and were led by a trained graduate student. After signing up for a time slot, participants were emailed a brief survey to collect demographic variables and obtain their consent. Upon entering the Zoom link, participants were given guidelines for the focus group and again asked to give written consent via the chat function. An undergraduate research assistant took notes, and Zoom’s transcription function was used for qualitative analysis of focus groups’ conversations.

Participants were asked what being “an activist” means to them and if they considered themselves to be an activist. Additionally, they were asked what behaviors they had personally employed that they would consider activist-like and how much commitment they considered these behaviors to entail. All behaviors mentioned across the four focus groups were considered, and overlapping items were consolidated. After conducting the fourth focus group, no new themes emerged, indicating that theoretical saturation had been reached. The result was a list of 21 items of activist-like behaviors (Table 1).

Table 1. Median perceived commitment, intention frequency, and IQR of behaviors in MTurk survey by activist type.

		Full Sample		Activist Only			Nonactivist Only
Item	Behavior	M	% Yes	M	% Yes	IQR	M	% Yes	IQR
1	Sign petition(s) to advocate for change	56.0	74.7	63.0*	74.3	45.0	45.0	75.1	67.0
2	Advocate for the issue on your person (e.g., wear a t-shirt, hat)	56.0*	66.7	59.0*	73.8	43.0	47.0*	58.7	56.0
3	Advocate for the issue with physical promotion materials (e.g., bumper sticker, sign in yard)	57.0*	63.9	61.0*	71.0	40.0	52.0*	56.1	55.0
4	Boycott a company or product	58.0	66.4	62.0	73.3	42.0	51.0*	58.7	53.0
5	Post about this issue on social media	59.0	72.4	62.0	78.6	38.0	54.0	65.6	61.5
6	Write a blog post	59.0	60.9	63.5	71.0	40.3	56.0*	49.7	47.0
7	Talk about the issue with someone who agrees with you about the issue	60.0	79.9	63.5*	75.7	40.0	52.0	84.7	60.8
8	Write a letter to your representative	61.0	66.7	62.0	77.1	34.0	61.0*	55.0	38.0
9	Hand out flyers with information on the issue	62.0*	60.9	62.0*	68.6	27.0	60.0*	52.4	41.8
10	Attend a protest	63.0	59.4	62.0	69.0	36.0	63.0*	48.7	38.8
11	Call your representative about the issue	63.0*	64.2	62.5*	71.4	34.0	63.5*	56.1	41.0
12	Donate money to organization	63.5	75.4	66.0	80.5	35.0	60.0	69.8	55.5
13	Educate yourself on this issue (e.g., read books, search online)	64.0*	52.6	64.0*	80.0	35.3	62.0*	87.8	50.0
14	Talk about the issue with someone who sees this issue differently than you do	65.0	76.4	67.0	80.5	33.0	61.0	72.0	44.0
15	Educate others on the issue	66.0	76.4	68.0*	73.3	30.0	65.0	79.9	39.5
16	Join a club and/or attend a meeting	66.0*	67.2	66.0*	77.6	32.0	65.0*	55.6	33.0
17	Speak in front of a crowd	69.0*	57.4	69.0*	71.9	33.3	70.0*	41.3	41.5
18	Become a volunteer for an organization	70.0	64.9	69.0	74.3	35.8	71.0*	54.5	36.5
19	Introduce new legislature	70.0*	60.2	67.0*	72.9	31.5	78.0*	46.0	43.0
20	Organize a fundraiser	71.0	55.1	68.0	66.7	37.3	76.0*	42.3	40.0
21	Organize a protest	72.5	52.9	70.0*	65.7	32.5	76.0	38.6	40.0

M stands for median. *Indicates item was kept in final scale. Scale organized by full sample median commitment low to high.

Results

Results from the focus groups revealed 21 distinct activist behaviors that were brought up in at least one focus group. Most behaviors (61.9%) were mentioned in two or more focus groups (Table A2)(Table A1). Participants mentioned a wide range of behaviors they had either done themselves or observed others doing such as attending protests, posting on social media, and calling a representative. There was sometimes debate about which activities required the highest commitment. Typically, more independent and private items (e.g., signing a petition, educating oneself) were mentioned as “easier,” whereas items that took more time or were more visible (e.g., organizing a protest, speaking in front of a crowd) were considered more “difficult.” For example, in focus group 1, participant 5 said: “People that are heavily involved in protests and who risk health and safety take on the most commitment.” Items that were mentioned in different focus group sessions but were highly similar (e.g., “attending a protest” and “going to a march”) were consolidated into one item for the survey.

Study 2

Design

To assess the scale items ascertained from the focus groups, a cross-sectional survey was conducted using Amazon’s Mechanical Turk (MTurk). The final analytic sample was N = 399 after dropping 19 participants for failing attention checks. Study design and procedures were deemed exempt by the university’s IRB.

Participants

Respondents were disproportionately white (67.7%), male (64.9%), and young (71.5% between the ages of 18 and 39). The sample consisted of mostly Democrats (57.6%; Republicans = 29.8%, Independents = 10.5%).¹ Participants were compensated $1.50.

Procedure

After providing consent, respondents were asked if they would describe themselves as an activist. Participants who indicated “yes” were asked to identify the social issue related to their activism behavior (open-ended) and then asked to categorize the cause (e.g., health, human rights). Similarly, those who indicated they were not an activist were instructed to indicate a social/political issue they cared about and subsequently categorize it. Next, for each of the 21 behaviors elicited from focus groups, respondents indicated if they would be willing to complete the behavior to promote positive change toward the social issue they cared about (yes/no). Then, they assessed the commitment level for each behavior. Finally, they completed the convergent and divergent scales and demographics.

Measures

Willingness to Participate

Respondents were asked to consider a social issue they care about and indicate which behaviors they would be willing to engage in to promote positive change. Twenty-one behaviors were listed, and respondents indicated yes (1) or no (0) for each one (M = 14.10, SD = 5.38).

Perceived Commitment

Participants indicated if they found each of the 21 activist behaviors to be easy and take little effort (low commitment) or to be difficult and take a lot of effort (high commitment). Responses were measured on a thermometer scale ranging from 0 to 100 (Table 1).

Activist Identity and Commitment Scale

Eight items measured activist identity on a seven-point scale (M = 4.60, SD = 1.78, Cronbach’s alpha, $\mathit{\alpha }$ = 0.97) from 1 “strongly disagree” to 7 “strongly agree.” Items included “being an activist is central to who I am” and “I take the time I need to engage in activism” (Klar & Kasser, 2009).

Activist Intention Scale

Four items measured activist behavioral intention on a seven-point scale (M = 5.03, SD = 1.33, Cronbach’s $\mathit{\alpha }$ = 0.86) from 1 “strongly disagree” to 7 “strongly agree.” Items included “I would join/belong to an organization that fights for my group’s political and legal rights” and “I would donate money to an organization that fights for my group’s political and legal rights” (Moskalenko & McCauley, 2009).

Trait Anger Scale

Trait anger was measured using 15 items on a four-point scale (M = 2.35, SD = 0.79, Cronbach’s $\mathit{\alpha }$ = 0.96) from 1 “never” to 4 “almost always.” Examples include “I have a fiery temper” and “I get angry when I’m slowed down by others’ mistakes” (Spielberger et al., 1983).

Dogmatism Scale

Eleven items measured dogmatism on a seven-point scale (M = 3.41, SD = 0.78, Cronbach’s $\mathit{\alpha }$ = 0.74) from 1 “strongly disagree” to 7 “strongly agree.” Examples include “People who disagree with me are usually wrong” and “I am a ‘my way or the highway’ type of person” (Infante & Rancer, 1982).

Results

Analysis plan

Thurstone scales are interval-level scales that are constructed to achieve equal-appearing intervals; making them distinctive from Likert scales where the interval is unknown (Tanujaya et al., 2022). Knowing the interval between items allows scholars to assess the level of commitment associated with certain behaviors. As such, when the final scale is constructed and used in studies, participants will indicate whether they would engage in a behavior (1) or not (0), and that response is weighted (multiplied) by its median. This study will determine the items with equal appearing intervals and their weight. Assuming that judgments of difficulty or commitment might be different for people who engage in activism relative to those who do not, we analyzed the data separately for activists (n = 210) and nonactivists (n = 189; Table 1). For each group, we ordered perceived commitment by the median. Then, consistent with Thurstone procedures (Edwards, 1957), we selected only one item from each median level. This procedure assists with achieving equally appearing intervals as well as not including redundant items. We retained items for the final scale by removing the item with a higher interquartile range (IQR). The IQR indicates variability across judges (in this case, the participants); it is important to choose items with the least variability to reduce noise. For example, for activists both joining a club and donating money had a median commitment score of 66.00. Since joining a club had a lower IQR (33.00) than donating money (55.50), the latter was dropped from the final scale.

Scale development

For activists, perceived commitment (median) ranged from 59 to 70. After removing nine items that shared medians and had the higher IQR (wide variance), we were left with 12 final items. CSAS scores for each participant were calculated by multiplying the item median by a 0 (if the participant said they would not do the behavior) or a 1 (if they said they would do the behavior) and averaging across the retained 12 behaviors (M = 47.63, SD = 14.46; range = 5.25 to 64.58). For nonactivists, perceived commitment (median) ranged from 47 to 78. After removing seven items that shared medians and had the higher IQR, we were left with 14 final items (M = 33.42; SD = 20.19; range = 0 to 62.54). To address RQ1, we examined items that were in both the 12-item activist and 14-item nonactivist scales, resulting in eight retained items that can be used for mixed groups of activists and nonactivists (M = 41.48; SD = 18.89; range = 0 to 63.38).

Tests of validity

Results of Pearson’s correlations found CSAS correlated strongly with both the activist identity scale (r = 0.52; p < .01) and activist intention scale (r = 0.57; p < .01), supporting H1. CSAS was also correlated with trait anger (r = 0.27; p < .01) and dogmatism (r = 0.19; p < .01). However, a series of Fischer’s z-tests indicate the correlation coefficients with the discriminant scales were significantly smaller (p < .001). Looking at the confidence intervals surrounding the correlations, the correlations between CSAS and the convergent scales do not overlap with the divergent scales (Table 2).

Table 2. Correlation matrix of all constructs comparing activists and nonactivists.

									Internal Reliability Coefficients (nonactivists)
	CSAS (all)	Lower Interval	Upper Interval	1	2	3	4	5	Cronbach’s α	McDonald’s ω
1. CSAS				1	.61**	.69**	.29**	.33**
2. Activist Identity and Commitment Scale	0.56**	0.48	0.62	.30**		.78**	.57**	.63**	0.98	0.99
3. Activist Intention Scale	0.60**	0.53	0.66	.33**	.77**		.35**	.50**	0.90	0.91
4. Dogmatism Scale	0.19**	0.09	0.28	−.07	−.02	−.10		.45**	0.75	0.84
5. Trait Anger Scale	0.27**	0.17	0.36	.00	.27**	.23**	.53**		0.96	0.97
Cronbach’s α (activists)					0.86	0.70	0.68	0.94
McDonald’s ω (activists)					0.89	0.77	0.87	0.95

**p < .01. The lower triangle shows the correlations for activists, while the upper triangle indicates the coefficients for nonactivists (refer to Table A4 for combined data). McDonald’s omega (ω; Hayes & Coutts, 2020) and Cronbach’s alpha (α) were used to measure scale reliabilities.

We also examined the correlations among all the scales, splitting the sample into activist and non-activist groups (RQ1; Table 2). For those participants who self-identified as activists, CSAS was not correlated with dogmatism or trait anger. For non-activists, the correlations are significant with small-to-medium effects (Gignac & Szodorai, 2016). These data reveal that CSAS can be discriminated from trait anger and dogmatism, particularly for self-identified activists.

Discussion

Social activism involves partaking in behaviors to try to bring about political or social change (Jang et al., 2021). The anger activism model (Turner, 2007) predicts that when anger is high and efficacy is strong, people will be more likely to engage in high-commitment activist behaviors. We argue that a Thurstone-style scale is ideal for measuring one’s willingness to partake in behaviors such as social activism, due to its ability to capture the escalating nature of the commitment as well as the complexities that go into behavioral intention (i.e., personal traits, finances) while maintaining unidimensionality.

Through our multistage approach, we developed 21 items aimed at measuring commitment to social activism. Ultimately, we narrowed the scale down to eight items that assess commitment for both activists and nonactivists, 12 items for just activists, and 14 for just nonactivists (Table 1). As expected, each subscale converged with activist identity and activism intention and could be discriminated from dogmatism and trait anger (Table 2).

Limitations of the study include focus groups being a college student sample. This may have impacted the activist behaviors that were listed, as age is related to activism engagement (Pfundmair et al., 2020). We attempted to address this issue by recruiting a more diverse population for Study 2 using an online platform to access and validate our scale. Additionally, our scale is firmly a scale of behavioral intention, which does not always align with behavior (Sheeran, 2002). Thus, we caution future researchers who use our scale to avoid making behavioral conclusions directly from only these measures. Work that incorporates CSAS while also measuring actual behavior is welcome.

We offer a methodological paper that contributes to a theoretical issue. The lack of proper measurement tools greatly inhibits our ability to accurately predict escalating political behavior and, as a result, our research suffers. Through the creation and validation of a new, escalating social activism scale, we proffer a novel approach to the study of social activism in communication research. In so doing, we facilitate the operationalization of social activism becoming harmonious with its conceptualization. Generally, this type of scale allows scholars to differentiate individuals who will engage in the minimum from those who will engage in time–consuming, effortful behaviors. Armed with this more detailed level of measurement, we can ask more complex theoretical questions and return clearer answers.

Open scholarship

This article has earned the Center for Open Science badges for Open Data and Open Materials through Open Practices Disclosure. The data and materials are openly accessible at https://osf.io/wxcza/?view_only=20755f3809994fb7b1ca2508590704c8.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Supplemental material

Supplemental data for this article can be accessed online at https://doi.org/10.1080/08824096.2024.2374413.

Additional information

Notes on contributors

Monique Mitchell Turner

Monique Mitchell Turner is Professor and Chair of the Department of Communication at Michigan State University. Her main focus is on the role that emotions play in the underlying psychosocial processes explaining message effects.

Rachel Barry Wade

Rachel Barry Wade is a doctoral candidate in the School of Communication at The Ohio State University. Her research focuses on the cognitive, metacognitive, and affective processes of persuasive message effects.

Ruth Heo

Ruth Heo is a doctoral student in the Department of Communication at Michigan State University. Her research interests lie in examining social influence in online contexts.

Qijia Ye

Qijia Ye is a doctoral student in the Annenberg School for Communication at the University of Pennsylvania. His research interests include persuasion, health communication, and communication technology.

Youjin Jang

Youjin Jang is a postdoctoral researcher at the Lineberger Comprehensive Cancer Center at the University of North Carolina at Chapel Hill. Her research interests lie at the intersection of persuasion and health and risk communication

Notes

1. For more information regarding participants in Study 1 and 2 refer to Tables A1 and A3 in the Appendix.

Appendix

Table A1. Demographic frequencies for focus groups.

Focus Groups
Gender	Frequency	%	Cumulative %
Female	12	52.2%	52.2%
Male	11	47.8%	100%
Age
18	6	26.1%	26.1%
19	3	13.0%	39.1%
20	6	26.1%	65.2%
21	2	8.7%	73.9%
22	4	17.4%	91.3%
23	2	8.7%	100.0%
Race
White	12	57.1%	57.1%
Hispanic/Latinx	5	23.8%	80.9%
African American/Black	2	9.5%	90.4%
Asian/Asian American/Pacific Islander	2	9.5%	100%

N = 23

Table A2. Frequency and perceived commitment of activist behaviors in focus groups.

	Behavior	Frequency	Perceived Commitment
1	Educate yourself on this issue (e.g., read books, search online)	1	Easy
2	Sign petition(s) to advocate for change	3	Easy
3	Post about this issue on social media	4	Easy
4	Donate money to organization	2	Easy
5	Write a letter to your representative	3	Moderate
6	Talk about the issue with someone who agrees with you about the issue	2	Easy
7	Talk about the issue with someone who sees this issue differently than you do	3	Moderate
8	Educate others on the issue	3	Moderate
9	Advocate for the issue on your person (e.g., wear a t-shirt, hat)	1	Moderate
10	Advocate for the issue with physical promotion materials (e.g., bumper sticker, sign in yard)	1	Moderate
11	Hand out flyers with information on the issue	2	Easy
12	Call your representative about the issue	1	Moderate
13	Attend a protest	4	Moderate
14	Boycott a company or product	1	Easy
15	Write a blog post	1	Moderate
16	Join a club and/or attend a meeting	3	Moderate
17	Become a volunteer for an organization	2	Moderate
18	Organize a fundraiser	1	Hard
19	Speak in front of a crowd	2	Hard
20	Organize a protest	4	Hard
21	Introduce new legislature	1	Hard

Frequency measured as number of focus groups in which item was mentioned.

Table A3. Demographic frequencies for MTurk survey by activist type.

	Activists		Nonactivists
	n	%	n	%
Gender
Male	141	67.1%	118	62.4%
Female	67	31.9%	68	36.0%
Non-binary	0	0.0%	1	0.5%
Age
18–29	62	29.8%	36	19.2%
30–39	92	44.2%	94	50.0%
40–49	28	13.5%	29	15.4%
50–59	21	10.1%	24	12.8%
60–69	5	2.4%	3	1.6%
70–79	0	0.0%	2	1.1%
Race
White	136	63.6%	134	70.2%
African American/Black	51	23.8%	23	12.0%
Asian/Asian American/Pacific Islander	11	5.1%	21	11.0%
Hispanic/Latinx	14	6.5%	8	4.2%
American Indian/Native Hawaiian	1	0.5%	4	2.1%
Other	1	0.5%	1	0.5%
Political Identity
Very liberal	37	17.6%	25	13.2%
Somewhat liberal	34	16.2%	54	28.6%
Moderate/middle of the road	25	11.9%	27	14.3%
Somewhat conservative	67	31.9%	58	30.7%
Very conservative	44	21.0%	25	12.7%
Political Party
Democratic	126	60.0%	104	55.0%
Republican	65	31.0%	54	28.6%
Independent	13	6.2%	29	15.3%
Other	4	2.0%	1	0.5%

N = 399

Table A4. Correlation matrix with entire sample.

	1	2	3	4	5
1. CSAS
2. Activist Identity	0.55**
3. Activist Intention	0.60**	0.80**
4. Dogmatism	0.19**	0.43**	0.25**
5. Trait Anger	0.27**	0.54**	0.44**	0.51**

**p < .01